Bearings Fault Detection Using Hidden Markov Models and Principal Component Analysis Enhanced Features

TL;DR

This paper introduces a method combining Hidden Markov Models and PCA-enhanced features to improve early fault detection in bearings, demonstrating promising results in a test bed environment.

Contribution

It presents a novel approach integrating HMM with PCA-enhanced features for more effective bearing fault detection.

Findings

Successful application on bearing test bed

Enhanced feature extraction improves fault detection accuracy

Method captures second order data structure

Abstract

Asset health monitoring continues to be of increasing importance on productivity, reliability, and cost reduction. Early Fault detection is a keystone of health management as part of the emerging Prognostics and Health Management (PHM) philosophy. This paper proposes a Hidden Markov Model (HMM) to assess the machine health degradation. using Principal Component Analysis (PCA) to enhance features extracted from vibration signals is considered. The enhanced features capture the second order structure of the data. The experimental results based on a bearing test bed show the plausibility of the proposed method.